English immersion, an instructional alternative that is popular among critics of bilingual education, has fared poorly in the U.S. Education Department’s first large-scale evaluation of the method, according to early results.
In the first year of a four-year longitudinal study by S.R.A. Technologies Inc., limited-English-proficient students in bilingual programs consistently outperformed “immersion strategy” students in reading, language-arts, and mathematics tests conducted in both English and Spanish.
This “unexpected” pattern was the opposite of what S.R.A. had predicted, according to David Ramirez, an employee of the contractor who is directing the federally funded study.
In a Dec. 19, 1985, summary of the test scores for the study’s advisory committee, however, Mr. Ramirez cautioned that the results are “tentative at best” and recommended that they be kept confidential to avoid “exacerbating an already volatile public debate.” A copy of the memorandum was obtained by Education Week.
Indeed, the findings are seen as likely to affect the debate over the Reagan Administration’s call for “flexibility” in serving the needs of LEP students. At the request of Secretary of Education William J. Bennett, the Congress is considering the removal of restrictions on federal funding for immersion and other alternatives to transitional bilingual education.
On seeing the Ramirez memo, one Republican member of the House called the test scores “dynamite,” sources said.
Especially perplexing to the S.R.A. researchers was the poor English-language performance of the immersion students, who had received the most English-language instruction. Moreover, the larger the native-language component of their schooling, the better the students performed in English.
The four-year study was designed to compare about 4,000 LEP kindergartners, 1st graders, and 3rd graders enrolled in immersion classes, “early-exit” or transitional bilingual education, and “late-exit” or bilingual maintenance programs. So far, the last group has shown the greatest overall progress in both languages of instruction.
Immersion--sometimes known as “sheltered English"--differs in important ways from “submersion,” also known as “sink or swim.” While subjects are taught in the second language, immersion instruction is geared to students’ level of English proficiency. Also, the teacher is often fluent in the students’ native language.
Transitional bilingual education--the predominant method supported by federal grants--is designed to prepare LEP children for mainstream classes, usually within three years. Maintenance programs, on the other hand, are aimed at preserving students’ bilingualism, and generally last six years or more.
Proof of immersion’s effectiveness among middle-class, “language majority” students in Canada has been widely accepted. But a controversy remains over the wisdom of transplanting this model to the United States to teach language-minority students of low socioeconomic status.
Proponents of bilingual education, while acknowledging that the S.R.A. study results are preliminary, have greeted the first-year scores as evidence that immersion is not the “panacea” for LEP students that some have claimed.
“There are no quick fixes,” said Ram6n L. Santiago, director of the Georgetown University Bilingual Education Service Center. “Politics won’t teach kids to read or do better in math. Policy must be informed by educational criteria.”
Rethinking Immersion
At the same time, the findings have prompted a prominent advocate of immersion strategies, Keith Baker of the Education Department’s office of planning, budget, and evaluation, to reconsider his views on second-language acquisition.
Extended exposure to English--a basic feature of the immersion approach--may “fatigue the learner so that learning becomes very difficult,” Mr. Baker said. Intervals of native-language instruction might help mitigate this effect and make learning “more efficient,” he added.
Mr. Baker conceded that this model closely resembles transitional bilingual education. Previously, on the basis of his 1981 review of evaluation research, which he conducted with Adriana de Kanter, he argued that “transitional bilingual education has had minimal success” and that “there is no empirical evidence to support its use.”
But the approach he now considers promising is used hardly anywhere in the United States, if at all, Mr. Baker said in a recent interview. Instead, he maintained, bilingual education typically features extended periods of instruction in each language--for example, English in the morning and Spanish in the afternoon.
“That shows Keith Baker hasn’t been in touch with the field or the language realities in the classroom,” contended James J. Lyons, legislative counsel of the National Association for Bilingual Education. A wide variety of approaches--including the one Mr. Baker described--are now in use in transitional bilingual programs, Mr. Lyons said.
The NABE official agreed that “learning fatigue” is an important problem for LEP students. “That’s what our people have been saying for 15 years and dealing with in the classroom,” Mr. Lyons said.
Some amount of native-language instruction is needed, he argued, although programs differ because children differ in their facility for learning languages. He added that considerable “flexibility” already exists for districts in meeting these needs.
External Factors
Mr. Baker qualified his “hypothesis” about the shortcomings of immersion by pointing to several external factors that may have lowered the immersion students’ performance in the S.R.A. study.
The 32 immersion classrooms being evaluated “are all new programs,” Mr. Baker said. “These programs may still be on their shakedown cruise. Another possibility is that the immersion programs were put in place in the most difficult schools--where there was the most severe English-language deficiency. The data suggest that this might be happening.”
A McAllen, Tex., immersion program in the sample “is located in the most heavily Hispanic and the most severely disadvantaged neighborhood” in the school district, he said. The children enrolled were less likely to have attended preschool classes than those in bilingual programs, he added.
In addition, Mr. Baker seconded the S.R.A. researcher’s concern that too much may be read into the first-year data.
“It’s like trying to call the winner of the Kentucky Derby based on the horse that reaches the first pole,” he said. Other assessments of immersion, such as those by Russell Gersten of the University of Oregon, have been more favorable, Mr. Baker said. Such variance in research findings, he added, remains an argument for greater flexibility in federal funding for services to LEP students, as advocated by Secretary Bennett.
Testing Hypotheses
The S.R.A. study was designed, according to Mr. Baker, to evaluate two conflicting theories of how LEP students learn:
- That maximizing English exposure will hasten English acquisition--"the time-on-task argument";
- That reading and other academic skills learned through the native language are “transferable,” an effect that aids second-language learning.
Mr. Baker argued that the first-year data contradict both these hypotheses. Although the late-exit bilingual students scored the highest, the transferability factor would not have had time to work, he said.
But in a paper delivered to NABE’s recent annual conference, Kenji Hakuta, a Yale University psychologist, argued that the transfer of literacy between languages has been well established through basic research into the relationship between bilingualism and cognition.
“Once the basic principles of reading are mastered in the home language, reading skills transfer quickly and easily to a second language,” Mr. Hakuta said.
This should be the preferred method, he added, “for children whose parents have little education and poor literacy skills . ... Such children run the serious risk of failing to learn to read if the problem of reading itself is made more difficult for them by being presented in a language they control poorly.”
In an interview, Mr. Hakuta said the poor results for immersion programs in the S.R.A. study, while preliminary, are “quite remarkable,” because the instructional alternatives being compared were carefully monitored for the amount of English used in the classroom. Frequently in evaluation studies, he said, program labels mask large instructional differences.
According to Mr. Ramirez of S.R.A., who explained his methodology at a NABE conference session, researchers determined that the immersion classes used English 90 percent of the time, compared with 67 percent in the early-exit bilingual programs and 33 percent in the late-exit bilingual programs.
Overall test scores from five school districts showed an inverse relation between English-language exposure and English-language proficiency among kindergartners and 1st graders.
In an interview, Mr. Ramirez argued that “it would be totally unprofessional and reprehensible” for educators or researchers to draw any conclusions from the first-year data.
Although the learning curves of the various groups of students “ran counter to what we predicted,” he said, “it could turn around next year.” No conclusions should be drawn until results of the study’s full four years are available, he argued.
According to Mr. Ramirez, the first-year data will be omitted from a forthcoming annual report on the project.
In addition to the 32 immersion classes, Mr. Baker said, 43 early-exit and 15 late-exit bilingual programs were included in the study, with more to be added next year. The late-exit students were tested only at the kindergarten and 3rd-grade levels.
By the time it is completed in 1988, the study is expected to cost $2.2 million, according to Edward Fuentes of the department’s office of bilingual education and minority-languages affairs.
